tatara-engine 0.2.243

Drivers, nix eval, cluster logic, P2P, and scheduling engine for tatara
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70

use anyhow::Result;
use std::sync::Arc;
use std::time::Duration;
use tracing::{debug, info, warn};

use crate::client::executor::Executor;
use crate::domain::evaluation::Evaluator;
use crate::domain::store_adapter::ClusterStoreAdapter;

/// Runs the scheduling loop: evaluates pending jobs and dispatches allocations.
///
/// The scheduler only proposes allocations when this node is the Raft leader.
/// This prevents duplicate allocations across a multi-node cluster.
pub struct Scheduler {
    evaluator: Evaluator,
    executor: Arc<Executor>,
    store: Arc<ClusterStoreAdapter>,
    eval_interval: Duration,
}

impl Scheduler {
    pub fn new(
        store: Arc<ClusterStoreAdapter>,
        executor: Arc<Executor>,
        eval_interval_secs: u64,
    ) -> Self {
        Self {
            evaluator: Evaluator::new(store.clone()),
            executor,
            store,
            eval_interval: Duration::from_secs(eval_interval_secs),
        }
    }

    /// Run the scheduler loop until cancelled.
    pub async fn run(&self) -> Result<()> {
        info!("Scheduler started (interval: {:?})", self.eval_interval);
        let mut interval = tokio::time::interval(self.eval_interval);

        loop {
            interval.tick().await;

            // Leader-affinity: only the Raft leader schedules new allocations.
            if !self.store.is_leader().await {
                debug!("Not leader, skipping scheduling tick");
                continue;
            }

            match self.evaluator.evaluate().await {
                Ok(allocations) => {
                    for alloc in allocations {
                        info!(
                            alloc_id = %alloc.id,
                            job_id = %alloc.job_id,
                            group = %alloc.group_name,
                            node = %alloc.node_id,
                            "Created allocation"
                        );
                        if let Err(e) = self.executor.start_allocation(alloc).await {
                            warn!(error = %e, "Failed to start allocation");
                        }
                    }
                }
                Err(e) => {
                    warn!(error = %e, "Evaluation cycle failed");
                }
            }
        }
    }
}